document_etl.md

Document ETL

ETL stands for "extract transform and load" and is the process of automating document processing for the purpose of putting specific information from documents into databases.

Word or Writer image extraction

(via unzip and emf conversion)

#!/bin/bash
# Neither of the websites provide a pixel perfect method.
# see also https://stackoverflow.com/questions/15063349/imagemagick-on-linux-to-convert-emf-to-png/28749719#28749719
# cited by https://askubuntu.com/questions/533665/convert-emf-to-png-image-using-command-line

suffix="_FILES"
folder_name="$1$suffix"
if [ -z "$1" ]; then
  echo "ERROR: nothing done since you must specify a docx file"
  exit 1
fi
if [ ! -d "$folder_name" ]; then
  if [ ! -f "$1" ]; then
    echo "ERROR: nothing done since $1 does not exist."
    exit 2
  fi
  unzip "$1"
fi
if [ ! -d "$folder_name" ]; then
  echo "ERROR: Nothing done since you must first extract $1 here to get $folder_name"
  exit 1
fi

ex_images_path="$folder_name/word/media"

echo "The code below is deprecated since it is not pixel perfect (is DPI based)"
echo "Instead, manually open "$ex_images_path/image*.emf files one by one in Inkscape, then in the program right-click the image that appears, then use 'Extract Image' feature."
exit 0


convert_emf()
{
  this_name="$1"
  source_path="$ex_images_path/$this_name.emf"
  dest_path="$this_name.png"
  #libreoffice --headless --convert-to png "$try_path"
  inkscape -e "$dest_path" --export-dpi=600 "$source_path"
}

try_convert_emf()
{
  this_name="$1"
  try_path="$ex_images_path/$this_name.emf"
  if [ -f "$try_path" ]; then
    echo "converting $this_name..."
    convert_emf "$this_name"
  else
    echo "no $try_path so quitting."
    exit 0
  fi
}

try_convert_emf "image1"
try_convert_emf "image2"
try_convert_emf "image3"
try_convert_emf "image4"
try_convert_emf "image5"
try_convert_emf "image6"
try_convert_emf "image7"
try_convert_emf "image8"
try_convert_emf "image9"
try_convert_emf "image10"